Community-Aware Vertex Ordering for Reference-Based Graph Compression: A Cross-Encoder Empirical Study
Researchers have developed a new vertex ordering method called Leiden+LLP for graph compression, which improves efficiency by analyzing community structures within the graph. This approach demonstrated significant savings, reducing bits per edge by 0.3 to 5.4 across various datasets and compression encoders. The study also introduced three new reference-based encoders (BG, CS, and CG) that offer further compression gains over existing methods, with the potential for low-overhead random access. AI
IMPACT Improves efficiency for graph compression techniques, potentially impacting data handling in AI systems.